Introducing Reliable Machine Learning with ALiX™

Current machine learning methods are highly sensitive to random numbers generated at the start of machine training. As a result, suboptimal selections of random parameters may cause unnecessarily long training times and poor prediction accuracy.

ALiX is the only system and method that reliably finds the best machine training solution without the need to select random parameters, improving confidence in machine answers and optimizing the machine learning processes to reduce cost, effort, and time to market.

Understand Your Data

ALiX is supervised machine learning. In this mode of AI, the “supervisor ” is a human expert that provides data with labels. The data may be clinical data, image data (such as MRIs and x-rays), family history, or other medical records, and the labels split the data into categories, such as “healthy” and “diseased,” or specify numeric properties of the data like dosage, growth rate, or pressure.

During the training process, ALiX learns how to fit the data to the labels. After training is complete, new data without labels can be examined, and ALiX will predict the values of the missing labels.

Increase Confidence and Reduce Costs

ALiX is a deterministic solution not available anywhere else. By using new and proprietary machine learning methods in the machine training process, customers can have confidence in repeatable results. For example, there is no need to perform a large number of training runs to hopefully find the best outcome. ALiX can find the best machine training solution in one training run, dramatically saving both human and computational costs, as well as time.

Scale Performance in the Data Center

ALiX is a uniquely parallel solution. Unlike other systems, ALiX can scale across multiple processors on multiple machines, and custom hardware acceleration is an option for an extra boost.